ARTICLE Disruption of POF1B Binding to Nonmuscle Actin Filaments Is Associated with Premature Ovarian Failure

Arnaud Lacombe,* Hane Lee,* Laila Zahed, Mahmoud Choucair, Jean-Marc Muller, Stanley F. Nelson, Wael Salameh, and Eric Vilain

Premature ovarian failure (POF) is characterized by elevated gonadotropins and amenorrhea in women aged !40 years. In a Lebanese family with five sisters who received the diagnosis of POF, we established linkage to the long arm of the X (between Xq21.1 and Xq21.3.3), using whole-genome SNP typing and homozygosity-by-descentmapping. By sequencing one candidate within that region, POF1B, we identified a point mutation localized in exon 10. This substitution of a nucleotide (GrA), at position 1123, results in an argininerglutamine mutation of the sequence at position 329 (mutation R329Q). All the affected family members were homozygous for the mutation, whereas the unaffected members were heterozygous. Because POF1B shares high homology with the tail portion of the human myosin, we assessed the ability of both wild-type and mutant POF1B to bind nonmuscle actin filaments in vitro. We found that the capacity of the mutant protein to bind nonmuscle actin filaments was diminished fourfold compared with the wild type, suggesting a function of POF1B in germ-cell division. Our study suggests that a homozygous point mutation in POF1B influences the pathogenesis of POF by altering POF1B binding to nonmuscle actin filaments.

A number of physiological changes are associated with ported on the long arm of the that spans menopause: decline in the number of ovarian follicles, Xq13.3-q26/27.10 Within this region, according to break- menstrual irregularities, ovarian hormonal deficiency, an- point mapping of X-autosome translocations, two regions ovulation, decreased fertility, and, finally, a complete and called “POF1” and “POF2” have been identified. The num- irreversible cessation of menses.1,2 In Western populations, ber of involved in POF is small and localized to it has been estimated that this phenomenon occurs at a either POF1 or POF2. POF1 extends from Xq21-qter and mean age of 50 years. However, in 2%–3% of women, is home to XPNPEP2 on Xq2511 and the FMR1 gene. POF2 cessation of menses occurs before age 40 years.3,4 This con- extends from Xq13.3 to Xq21.1 and harbors DACH2 in dition, called “premature ovarian failure” (POF),5 is char- Xq21.3,12 DIAPH2 in proximal Xq22,13 and POF1B, the acterized by primary or secondary amenorrhea, hypoes- gene that is the subject of the present study. Individuals trogenism, and elevated serum-gonadotropin levels. with the FMR1 premutation have idiopathic POF. The etiology of POF is poorly understood because its POF1B is thus located within the critical region for nor- known causes are highly heterogeneous. POF is most com- mal ovarian function, is known to escape X inactivation, monly caused by the presence of autoantibodies,6 but the and was identified as a candidate gene for POF in a patient disorder can also have genetic origins.7,8 Although the X with a breakpoint in the third intron because of an X;1 chromosome undergoes inactivation in somatic tissues, it translocation.14,15 The function of the POF1B protein re- is reactivated in oogonia, and both alleles of the X-chro- mains unknown, and there is no genetic evidence of its mosome genes involved in ovarian morphogenesis are role in POF. A mutation analysis of POF1B performed in required for normal ovarian development. As a result, X an Italian population showed that a number of heterozy- monosomy or X rearrangements disrupting genes of the gous sequence variants were present in both patients with X are sufficient to cause ovarian failure. Au- POF and healthy controls. It was concluded that the POF1B tosomal mutations in the follicle-stimulating hormone gene was not associated with POF.16 (FSH) and in both FSH and luteinizing-hormone (LH) re- In the present study, we describe a Lebanese familial case ceptor genes have already been reported.9 of POF, with linkage to Xq21. We identify a point muta- It has often been observed that abnormalities in the X tion in the candidate gene POF1B and demonstrate that chromosome are present in women affected by POF. A this mutation disrupts the normal binding of POF1B to “critical region” for normal ovarian function has been re- nonmuscle actin filaments.

From the Departments of Human Genetics (A.L.; H.L.; S.F.N.; E.V.), Pediatrics (E.V.), Urology (E.V.), and Psychiatry and Biobehavioral Science (S.F.N.), University of California at Los Angeles; Departments of Pathology and Laboratory Medicine (L.Z.) and Division of Endocrinology, Department of Internal Medicine (M.C.), American University of Beirut Medical Center, Beirut; Institut de Physiologie et Biologie Cellulaire, Centre National de la Recherche Scientifique–Unite´ mixte de Recherche 6187–Poˆle Biologie Sante´, Poitiers, France (A.L.; J.-M.M.); and Division of Endocrinology, Department of Medicine, Harbor-UCLA Medical Center and Los Angeles Biomedical Research Institute, Torrance, CA (W.S.) Received January 13, 2006; accepted for publication April 21, 2006; electronically published May 26, 2006. Address for correspondence and reprints: Dr. Eric Vilain, David Geffen School of Medicine at UCLA–Gonda Center, Room 5506, 695 Charles Young Drive South, Los Angeles, CA 90095-7088. E-mail: [email protected] * These two authors contributed equally to this work. Am. J. Hum. Genet. 2006;79:113–119. ᭧ 2006 by The American Society of Human Genetics. All rights reserved. 0002-9297/2006/7901-0012$15.00

www.ajhg.org The American Journal of Human Genetics Volume 79 July 2006 113 Material and Methods (Multipoint Engine for Rapid Likelihood Interference) and MINX Pedigree (Merlin for X chromosome). MAPMAKER/HOMOZ version 0.9 was used to perform homozygosity-mapping analysis of 11,149 SNPs We report a familial Lebanese case of POF wherein the proband’s from the 10K mapping array. Multipoint LOD score was calcu- parents, who are consanguineous, had five affected daughters, lated with MAPMAKER/HOMOZ for the region flanking the au- one healthy son, four other sons (two with albinism and blind- tozygous segment of interest. Because of the program constraints, ness due to retinitis pigmentosa, one with familial Mediterranean the genome scan was run using a 50-marker window. Asian allele fever/Behcet disease [FMF/Behcet], and one who died at age 14 frequencies provided by Affymetrix were used for the analysis. years), and one healthy daughter (fig. 1A). The fertile, unaffected Haplotypes were constructed with Merlin, and a list of candidate daughter is also in a consanguineous marriage and has several genes was generated using University of California–Santa Cruz affected children with blindness due to retinitis pigmentosa and/ (UCSC) resources from the genomic region identified by the link- or with FMF/Behcet. We had access to the DNA of the healthy age result. There are 270 known genes in this region. son, all the daughters, and the mother but not the father of all the affected daughters, who was deceased, and the three sons Sequencing with albinism/blindness or FMF/Behcet, who declined to be inter- viewed for the study. All three of these individuals were married From the list of candidate genes, we performed a mutation screen- and had progeny, but we have little clinical information available ing of diaphanous (MIM 300108) and POF1B (MIM 311360), two beyond what was given from participating family members. genes known to be interrupted by a breakpoint in patients af- fected by POF. Primers were designed for each exon and then POF Diagnosis were amplified by PCR. PCR product was sequenced by an auto- mated DNA sequence analyzer (ABI Prism Tm 373 [Perkin-Elmer]). The proband (individual 41) was aged 31 years at presentation. Finally, the sequences were analyzed and screened for mutation. She is single, has never been pregnant, and had a history of de- layed puberty with primary amenorrhea. She denied any history Incidence of the R329Q Variant in the Lebanese Population of galactorrhea, anosmia, or headaches. She continued to increase in height until she began sex-hormone treatment. Relevant as- Presence of the R329Q variant was investigated in 92 women from pects of her physical exam showed a height of 181 cm and weight the same ethnic background as our family. Individuals were known of 81 kg, normal axillary hair, and breasts of Tanner stage 3. She to be fertile and had a normal obstetric history; therefore, they also has documented osteoporosis and “weak teeth.” She had five were appropriate controls for our study. The DNA was collected sisters who were aged 45, 41, 39, 34, and 29 years at the time of in an anonymous way, with complete deidentification. Mutation her diagnosis. The eldest was married to a first cousin with whom screening in exon 10 of each control was performed using a di- she had children; among her children are two daughters who are deoxy sequencing method (Agencourt Bioscience). both blind due to retinitis pigmentosa and who have normal menses, LH, and FSH. The latter four sisters also have primary Protein Domain Prediction amenorrhea; in at least one sister, osteoporosis and dental prob- lems similar to those of the proband are documented. One of the To assess POF1B protein function, the protein sequence was sub- proband’s albino brothers also married his first cousin and has a mitted to different database sources: National Center for Bio- daughter who is infertile, but no further clinical information on technology Information (NCBI) protein alignment, NCBI protein her was available. The second albino brother married an unrelated domain prediction, and Predict Protein software. partner and has two daughters who have normal menses. The proband was treated with Premarin and Provera, which resulted Plasmids in resumption of her menses and advancement of her breasts and pubic hair to Tanner stage 5 after 2 years of therapy. Her electro- POF1B cDNA was purchased from OriGene. To produce the cDNA lytes were within normal limits. Before treatment with sex hor- mutant in vitro, we used the QuikChange Site-Directed Muta- mones, her LH was 11.5 IU/liter, FSH was 56.7 IU/liter, E2 was genesis Kit (Stratagene) and followed the manufacturer’s instruc- 8.9 IU/liter, and testosterone was 39 IU/liter. Her karyotype was tions. The mutated oligonucleotide primers were designed as fol-  46,XX. Adrenocorticotropic hormone (ACTH)-stimulation test lows: forward, 5 -GCTAAATGTGGACAGCACTAGTTGGAGTGA-   showed dehydroepiandrosterone sulfate of 2,009 IU/liter and CTTATCAG-3 ; reverse, 5 -CTGATAAGTCACTCCAACTAGTGCT-  17OH-P of 0.35 at baseline, which went to 2,363 and 3.15, re- GTCCACATTTAGC-3 . Sequences and presence of the point mu- spectively, 60 min after ACTH stimulation. tation were checked by automated DNA sequencing (ABI Prism TSm 373 [Perkin-Elmer]). Linkage In Vitro Protein Production After obtaining informed consent from all study individuals, blood samples were collected and genomic DNA was extracted by stan- Both wild-type and mutant cDNA were used as a template to dard salt-precipitation technique.17 The research protocol was ap- produce proteins in vitro, with use of a transcription translation proved by the local institutional review board. Linkage was tested kit (TNT Quick Coupled Transcription/Translation kit [Promega]). using a homozygosity-by-descent mapping approach, with data Yields of protein were estimated from the incorporation of 35S- derived from the GeneChip Mapping 10K array (Affymetrix). To methionine in reactions that were performed in parallel with non- test for linkage, SNP calls generated by GeneChip DNA Analysis radiolabeled syntheses. The 35S-radiolabeled wild-type and mu- Software were converted to different formats by scripts written tant POF1B proteins were run on a 12% SDS-PAGE gel. The gel in the lab for each program used (available from H.L. on request). was dried, exposed, and developed using a phosphorimager Mendelian and non-Mendelian errors were detected with Merlin cassette.

114 The American Journal of Human Genetics Volume 79 July 2006 www.ajhg.org Figure 1. Pedigree and linkage. A, Pedigree. B, Haplotype structure of the pedigree. Homozygous blocks on the X chromosome that are shared by all five affected daughters are shown as blackened bars. Each marker is labeled with its name, and the genetic distance appears to the right; blackened symbols indicate affected individuals. Fifty SNPs in the middle of each block were omitted because of the space constraint. Note that all affected daughters share identical haplotypes over a 21-Mb homozygous interval that the healthy daughter does not share. C, Genomewide-scan linkage analysis under an assumption of recessive inheritance and first-cousin consan- guineous mating. Multipoint LOD scores are plotted along the Y-axis, and chromosomal position is plotted along the X-axis. Chromosomes are listed in numerical order across the top. Actin-Myosin Binding Study chromosomes, we scanned the remaining genome with MAPMAKER/HOMOZ for homozygosity-mapping analy- The ability of both wild-type and mutant POF1B protein to bind nonmuscle actin filaments in vitro was assessed using a non- sis. Assuming the parents are first cousins, we performed muscle actin kit purchased from Cytoskeleton. A constant and analysis under an autosomal recessive inheritance model. equal quantity of wild-type and mutant stock protein were in- Throughout the autosomal chromosomes, there was no cubated with nonmuscle actin at room temperature for 30 min. homozygous stretch shared by only the affected daugh- Each experiment was run in duplicate and also with a control in ters and not by the unaffected daughter. The locus of link- which actin was substituted by an identical volume of actin buffer. age was localized to the long arm of the X chromosome Ultracentrifugation was performed at150,000 g for 1.5 h at 4ЊC. (genomic region between Xq21.1 and Xq 21.3.3), corre- Both supernatants and pellets were run on a 4%–20% (w/v) SDS- sponding to a 21-Mb region (fig. 1C). Because of the long PAGE gel. The gel was dried, exposed, and developed using a length, this interval is likely autozygous in all of the af- phosphorimager cassette. fected daughters. On the basis of the physical interval, a list of 270 known genes was generated, and two candidate Results genes (POF1B and diaphanous) were analyzed. Linkage A panel of 110,000 SNPs was typed on all individuals with Family Members with POF Who Carry a Point Mutation use of the 10K-SNP mapping assay (Affymetrix). The data in Exon 10 of POF1B from this microarray-based approach resulted in high in- To identify a possible involvement for one of the candi- formation content throughout the genome. The mean call dates in POF, we screened each exon by direct sequencing –with a range of 89.6%) 0.03% ע SD) was94%ע) rate of two candidates that have been demonstrated to be dis- 96.63%). The distribution of genotype calls was consistent rupted by a breakpoint (localized in introns) in patients over AA, AB, and BB calls, showing slightly lower AB calls affected by POF: diaphanous and POF1B. in most of the samples, as is expected for an inbred family. Sequencing of diaphanous exons failed to reveal a mu- Mendelian error rate and non-Mendelian error rate were tation for any of the family members (data not shown). 0.08% and 0.16%, respectively. Genotypes for SNPs with However, when screening POF1B, we identified a point at least one error were changed to “No Call” across all the mutation in exon 10, a GrA substitution at np 1132 (fig. family members before starting the analysis. Since there 2). This variant results in an amino acid change of argi- was no information on the genetic relationship between ninerglutamine at position 329 (R329Q) in the protein the parents of the affected women, we counted the pos- sible autozygous segments in the autosomal chromosomes of the descendants, to determine whether it was likely that the affected daughters were from a consanguineous mat- ing, which would permit us to use homozygosity by de- scent to map the disease interval.18 Identification of the autozygous segments by counting the long homozygous segments was possible because of the high-density SNPs with mean genetic map distance of 0.32 cM and mean heterozygosity of 0.37 cM. Since regions of autozygosity are expected to be large with first- or second-cousin in- breeding,18 we counted the number of segments with a min- imum of 10 consecutive homozygous SNPs that spanned at least 10 cM in all autosomes. The estimated number of autozygous blocks 110 cM is 6–10 for first-cousin mating, 3–5 for first-cousin-once-removed mating, and 1–3 for second-cousin mating. The six siblings in our study have homozygous blocks, which 2.7 ע SD) of7ע) an average strongly suggests that they are descendents of a consan- guineous marriage of recent relatives (first-cousin-once- removed marriage or first-cousin marriage). We searched through the genome and detected a single Figure 2. Representative sequence chromatogram of exon 10 of the POF1B gene after sequencing of the members of the family. long segment of homozygosity on the X chromosome that A, Sequence analysis of the mother’s and unaffected daughter’s was shared only by the affected daughters (fig. 1B). The DNA. At np 1132, those members carry either a cytosine (C) or a unaffected daughter was heterozygous for that region. Hap- thymine (T), which shows they are heterozygous. B, Sequence an- lotypes were constructed with Merlin, and the homozy- alysis of an affected daughter’s DNA. At the same np 1132, all gous segment spanned bases 41926093–92958387 (UCSC patients carry a T and are homozygous for the mutation. C, Se- May 2004 assembly). To rule out the possibility that there quence analysis of DNA of an unrelated female control. The control could be shared autozygous segments on the autosomal individual is a homozygote and carries a C.

116 The American Journal of Human Genetics Volume 79 July 2006 www.ajhg.org sequence. The unaffected women screened in the family (one daughter and the mother) are both heterozygous at this position, whereas all five affected daughters are homo- zygous for the mutation. Since the father died before the time of investigation, we were unable to screen his DNA. It is conceivable that the variant detected is a common SNP. To measure the incidence of the R329Q variant in the general Lebanese population, we tested whether the nucleotide variant is present in 92 individuals from a sim- ilar ethnic background. Although none carried the muta- tion in a homozygous state, results of sequence screening (data not shown) revealed that the mutation was present, in a heterozygous state, in 4 of the 92 control women. Thus, the estimated allele frequency in Lebanon is 2.2%. Figure 3. Nonmuscle actin–binding assay of wild-type (WT) and 35 The predicted rate of homozygosity in this population mutant POF1B. A, A constant and equal concentration of S-ra- would be 0.048% and could account for a fraction of the diolabeled wild-type and mutant POF1B was incubated with 12 mM nonmuscle actin and was spun down at150,000 g for 1.5 h at POF detected in this population. -4؇C. As a control, the same concentrations of 35S-radiolabeled pro teins were incubated under the same conditions, but nonmuscle POF1B Protein Shares a High Degree of Homology actin was replaced with the same volume of actin buffer. The top with Human Myosin panel represents SDS-PAGE gel picture after 70-h exposure from su- Although POF1B was identified as a candidate for POF,15 pernatant of each sample run in duplicate (Supernatant), whereas the bottom panel corresponds to SDS-PAGE gel picture from pellets little is known about its expression and potential role. To (Pellet) after the same time of exposure. Lanes correspond to wild- begin investigation of its function, we performed protein- type POF1B (1 and 2) or mutant POF1B (3 and 4), in the presence sequence comparisons against different databases. Results (2 and 4) or absence (1 and 3) of nonmuscle actin filaments. from NCBI protein alignment indicate that the protein Results are representative of three independent experiments. B, shares 42% similarity and 17% identity with the human SDS-PAGE analysis of in vitro–translated proteins. The same amount perinatal myosin, also named “heavy polypeptide 8.” of in vitro–translated protein reaction of wild-type and mutant This identity localizes to the C terminus and is, in fact, POF1B was run on an SDS-PAGE gel, to measure the yield of pro- a myosin tail. No similarity with other proteins was found duced proteins. C, Quantification of the radioactive signal in the for sequence at the N-terminus. pellet after the nonmuscle actin–binding assay. A triple asterisk ! A “predict protein” search indicated that the variant alters (***) indicatesP .005 . the arginine with an amino acid triplet, serine-leucine- p arginine, which is part of a protein kinase C (PKC) phos- the R329Q variant (P .0024 ) (fig. 3C). There was no dif- phorylation-recognition site. Thus, this rare variant may ference in the yield of protein produced after the in vitro have functional consequences. translation (fig. 3B).

Mutant POF1B Binds Nonmuscle Actin with Less Affinity Than Wild-Type POF1B In Vitro Discussion Previous studies have shown that, during ovarian devel- opment, the alteration of the final stock of oocytes could lead to POF. Given these findings and the similarity of the The molecular mechanisms that lead to POF are poorly C terminus of POF1B to the myosin tail of the human understood. Most frequently, POF cases are idiopathic or myosin, we next tested the hypothesis that POF1B could sporadic.3 Previous studies have described familial cases of interact with actin filaments and therefore have a possible POF with a dominant mutation or an association with X- role in the cell-division process. To do so, we assessed the chromosome abnormalities.19,20 A major obstacle to the ability of both the wild-type and mutant proteins to as- identification of new genes responsible for POF is the lack semble with nonmuscle actin in vitro (fig. 3). We show of familial cases with full pedigrees.21 In the present study, that, in the supernatant fraction, which mainly contains we show that a homozygous point mutation in POF1B, a protein not bound to actin, the decrease of intensity of gene mapping to the distal end of the POF2 critical in- the signal after actin binding in wild-type POF1B is larger terval on X chromosome, may be linked to the pathogen- than in R329Q mutant (fig. 3A). Conversely, in the pellet, esis of POF. We show that each affected 46,XX daughter in the presence of actin, the signal was more intense in in a Lebanese family carries a homozygous R329Q variant the wild type than in R329Q mutant (fig. 3A). After quan- in exon 10 of POF1B at np 1132, whereas unaffected fe- tification of three independent experiments (each in du- males (one healthy daughter and the mother) are hetero- plicate), we recorded an average 48% binding to actin fil- zygous for this variant. This variant alters the binding of ament with the wild-type POF1B, whereas it was 12% with the POF1B protein to nonmuscle actin.

www.ajhg.org The American Journal of Human Genetics Volume 79 July 2006 117 POF1B and POF actin filaments with less affinity than the wild type, we assessed the capacity of both wild-type POF1B and the POF1B, a gene found only in vertebrates, was discovered mutant to bind nonmuscle-actin filaments in vitro. Our elsewhere to be interrupted by an X;1 balanced transloca- data demonstrated that the variant POF1B binds nonmuscle tion in a patient with secondary amenorrhea.15 Recently, actin with less affinity than the wild type, suggesting that a mutation analysis study performed on Italian patients the R329Q variant affects the binding of POF1B with actin. with POF showed that, among several polymorphisms, 1 of 223 patients carried the heterozygous R329Q muta- Molecular Mechanisms of POF1B in POF tion in the POF1B gene, which was also found in 3 of 900 healthy controls. The authors concluded that there was Between E16.5 and P5 (during the POF1B expression time no significant association between POF1B and POF.16 The frame), while germ cells within germline cysts enter mei- lack of an association in the Italian study could be due osis and progress through the stages of pachytene and either to population differences or to the fact that most diplotene, massive germ-cell apoptosis occurs.29 In mice, of the Italian patients had secondary amenorrhea, whereas as in humans, this natural process of female germ-cell loss our patients had primary amenorrhea. Therefore, the eti- occurs to eliminate abnormal cells.30–32 The disruption of ology of each type is likely to be different. POF1B binding to actin filaments may have various con- Our data suggest that, in a homozygous state, R329Q is sequences for ovarian-follicle formation. The reported in- associated with POF. Since the mother and one daughter, terval of expression of the POF1B mRNA in mouse pre- both unaffected by POF, were heterozygous, and since there cludes its involvement in primordial-germ-cell migration, could be consanguinity in this family, a recessive mode since it is first detected at day 16.5 post coitum. Further- of inheritance is likely. In the control group—women from more, at this particular time interval, synaptonemal-com- the same ethnicity as our familial case—we identified four plex assembly has occurred, and oocytes are in the pachy- heterozygous mutations in 92 individuals. This high fre- tene stage. The interval of expression is consistent with a quency of this genotype in Lebanon increases the likeli- role of the protein in the transition from pachytene to hood of the presence of R329Q at the homozygous state. diplotene and dictyate arrest. The period from day 16.5 to Although there is no public health information about the birth is characterized by massive germ-cell attrition incidence of POF in the Lebanese population available at through programmed cell death, due to either failure of this time, we speculate that there is a relatively high rate completion of meiosis 1 or premature initiation of met- of the condition in that country. Sporadic and familial aphase 1.33 On the basis of the above and the homology cases should be investigated, in a large number of Leba- of POF1B to myosin, we speculate that POF1B could play nese women, for the presence of R329Q variants. a role in the pairing of meiotic chromosomes and that alteration in its function could lead to cell death and a POF1B Protein Function drastic reduction in the final number of oocytes created during ovarian development. Although POF1B is one of the candidate genes for POF, its Alternatively, POF1B could be involved in the regulation function remains unknown. Protein-function predictions of germ-cell apoptosis through a role in cytoskeletal dy- suggest that POF1B shares homology with the myosin tail namics. Alterations in the actin cytoskeleton have been portion of the human myosin protein. Also, predictions recently implicated in release of reactive oxygen species support the hypothesis that the presence of the R329Q from mitochondria and germ-cell apoptosis.34 POF1B may mutation affects a phosphorylation site represented by a act as an antiapoptosis factor, to slow down the process triplet of amino acids: serine-leucine-arginine, normally of germ-cell loss. As a consequence, loss of function of recognized by the PKC. POF1B could lead to exaggerated germ-cell apoptosis and Myosin plays an important role in cytokinesis. At the POF. The precise molecular mechanisms of action of POF1B time of division, filaments of actin and myosin form a in POF will be elucidated by animal models targeting POF1B contractile ring, a structure that is fundamental and re- in ovarian germ cells. quired for normal cell division.22–27 Also, it is well known that phosphorylation events are essential for myosin pro- tein function. In mammalian smooth muscle and non- Acknowledgments muscle systems, myosin ATPase activity, which regulates We thank Dr. Fleming for her thoughtful suggestions. Microarray the motor function of the protein, is highly dependent studies were performed within the UCLA DNA Microarray Facil- 28 on phosphorylation events. Given the high degree of ity. We thank Zugen Chen, for help in array performance, and homology of POF1B and the myosin tail of the human Allen Day, for computational support. myosin protein, we speculate that POF1B is a protein in- volved in actin-filaments interaction and that the R329Q Web Resources variant leads to a lack of phosphorylation at the serine- The URL for data presented herein is as follows: leucine-arginine site, which causes the loss of function of the POF1B protein. Online Mendelian Inheritance in Man (OMIM), http://www.ncbi To test the hypothesis that the POF1B mutant binds .nlm.nih.gov/Omim/ (for diaphanous and POF1B)

118 The American Journal of Human Genetics Volume 79 July 2006 www.ajhg.org References for premature ovarian failure, DACH2 and POF1B. Hum Re- prod 19:2759–2766 1. McKinlay S, Jefferys M, Thompson B (1972) An investigation 17. Miller SA, Dykes DD, Polesky HF (1988) A simple salting out of the age at menopause. J Biosoc Sci 4:161–173 procedure for extracting DNA from human nucleated cells. 2. Wyshak G, Frisch RE (1982) Evidence for a secular trend in Nucleic Acids Res 16:1215 age of menarche. N Engl J Med 306:1033–1035 18. Lander ES, Botstein D (1987) Homozygosity mapping: a way 3. Coulam CB, Adamson SC, Annegers JF (1986) Incidence of to map human recessive traits with the DNA of inbred chil- premature ovarian failure. Obstet Gynecol 67:604–606 dren. Science 236:1567–1570 4. Vegetti W, Marozzi A, Manfredini E, Testa G, Alagna F, Ni- 19. Sala C, Arrigo G, Torri G, Martinazzi F, Riva P, Larizza L, Phi- colosi A, Caliari I, Taborelli M, Tibiletti MG, Dalpra L, Cro- lippe C, Jonveaux P, Sloan F, Labella T, Toniolo D (1997) signani PG (2000) Premature ovarian failure. Mol Cell En- Eleven X chromosome breakpoints associated with prema- docrinol 161:53–57 ture ovarian failure (POF) map to a 15-Mb YAC contig span- 5. de Moraes-Ruehsen M, Jones GS (1967) Premature ovarian ning Xq21. Genomics 40:123–131 failure. Fertil Steril 18:440–461 20. Vegetti W, Grazia Tibiletti M, Testa G, de Lauretis Y, Alagna 6. Wheatcroft N, Weetman AP (1997) Is premature ovarian fail- F, Castoldi E, Taborelli M, Motta T, Bolis PF, Dalpra L, Cro- ure an autoimmune disease? Autoimmunity 25:157–165 signani PG (1998) Inheritance in idiopathic premature ovar- 7. Bione S, Toniolo D (2000) X chromosome genes and pre- ian failure: analysis of 71 cases. Hum Reprod 13:1796–1800 mature ovarian failure. Semin Reprod Med 18:51–57 21. Simpson JL, Rajkovic A (1999) Ovarian differentiation and 8. Santoro N (2001) Research on the mechanisms of premature gonadal failure. Am J Med Genet 89:186–200 ovarian failure. J Soc Gynecol Investig 8:S10–S12 22. Glotzer M (1997) Cytokinesis. Curr Biol 7:R274–R276 9. Christin-Maitre S, Vasseur C, Portnoi MF, Bouchard P (1998) 23. Glotzer M (1997) The mechanism and control of cytokinesis. Genes and premature ovarian failure. Mol Cell Endocrinol Curr Opin Cell Biol 9:815–823 145:75–80 24. Mabuchi I (1986) Biochemical aspects of cytokinesis. Int Rev 10. Therman E, Laxova R, Susman B (1990) The critical region Cytol 101:175–213 on the human Xq. Hum Genet 85:455–461 25. Mabuchi I, Okuno M (1977) The effect of myosin antibody 11. Prueitt RL, Ross JL, Zinn AR (2000) Physical mapping of nine on the division of starfish blastomeres. J Cell Biol 74:251– Xq translocation breakpoints and identification of XPNPEP2 263 as a premature ovarian failure candidate gene. Cytogenet Cell 26. Robinson DN, Spudich JA (2000) Dynacortin, a genetic link Genet 89:44–50 between equatorial contractility and global shape control dis- 12. Prueitt RL, Chen H, Barnes RI, Zinn AR (2002) Most X;au- covered by library complementation of a Dictyostelium dis- tosome translocations associated with premature ovarian fail- coideum cytokinesis mutant. J Cell Biol 150:823–838 ure do not interrupt X-linked genes. Cytogenet Genome Res 27. Robinson DN, Spudich JA (2000) Towards a molecular un- 97:32–38 derstanding of cytokinesis. Trends Cell Biol 10:228–237 13. Bione S, Sala C, Manzini C, Arrigo G, Zuffardi O, Banfi S, 28. Tan JL, Ravid S, Spudich JA (1992) Control of nonmuscle Borsani G, Jonveaux P, Philippe C, Zuccotti M, Ballabio A, myosins by phosphorylation. Annu Rev Biochem 61:721–759 Toniolo D (1998) A human homologue of the Drosophila me- 29. Pepling ME, Spradling AC (2001) Mouse ovarian germ cell lanogaster diaphanous gene is disrupted in a patient with pre- cysts undergo programmed breakdown to form primordial mature ovarian failure: evidence for conserved function in follicles. Dev Biol 234:339–351 oogenesis and implications for human sterility. Am J Hum 30. Baker TG (1972) Gametogenesis. Acta Endocrinol Suppl (Co- Genet 62:533–541 penh) 166:18–41 14. Gersak K, Meden-Vrtovec H, Peterlin B (2003) Fragile X pre- 31. Krakauer DC, Mira A (1999) Mitochondria and germ-cell death. mutation in women with sporadic premature ovarian failure Nature 400:125–126 in Slovenia. Hum Reprod 18:1637–1640 32. Morita Y, Tilly JL (1999) Oocyte apoptosis: like sand through 15. Riva P, Magnani I, Fuhrmann Conti AM, Gelli D, Sala C, an hourglass. Dev Biol 213:1–17 Toniolo D, Larizza L (1996) FISH characterization of the Xq21 33. De Felici M, Klinger FG, Farini D, Scaldaferri ML, Iona S, Lo- breakpoint in a translocation carrier with premature ovarian bascio M (2005) Establishment of oocyte population in the failure. Clin Genet 50:267–269 fetal ovary: primordial germ cell proliferation and oocyte pro- 16. Bione S, Rizzolio F, Sala C, Ricotti R, Goegan M, Manzini MC, grammed cell death. Reprod Biomed Online 10:182–191 Battaglia R, Marozzi A, Vegetti W, Dalpra L, Crosignani PG, 34. Gourlay CW, Ayscough KR (2005) The actin cytoskeleton: a Ginelli E, Nappi R, Bernabini S, Bruni V, Torricelli F, Zuffardi key regulator of apoptosis and ageing? Nat Rev Mol Cell Biol O, Toniolo D (2004) Mutation analysis of two candidate genes 6:583–589

www.ajhg.org The American Journal of Human Genetics Volume 79 July 2006 119